Re: [RFC PATCH] edac: add support for ARM PL310 L2 cache parity

[Punnaiah Choudary]

There is a driver file cache-l2x0.c under arch/arm/mm for pl310 cache
configuration and management. Russel king had suggested to use
single driver file for both pl310 edac implementation and cache-l2x0.c
Here is the thread
http://www.spinics.net/lists/arm-kernel/msg320407.html

please provide your inputs

Thanks,
Punnaiah

On Wed, Apr 9, 2014 at 8:49 PM, Borislav Petkov <bp@xxxxxxxxx> wrote:
> On Wed, Apr 09, 2014 at 08:18:28AM -0500, Rob Herring wrote:
>> I don't think so, the PL310 is present on lots of ARM chips besides
>> Xilinx. I don't know how many support parity as that is optional. In
>> fact the highbank_l2_edac.c is for the PL310 as well, but the
>> registers it uses is all custom logic added for ECC and there is no
>> part of the PL310 h/w used by the driver.
>
> Oh ok, so highbank_l2 and PL310 could theoretically be merged together
> in one compilation unit, even if they don't really share code at all...
>
>> If there is lots duplication, then that's a sign the framework needs
>> to handle more of the boilerplate pieces. There could be a "simple"
>> driver/library for devices which are no more than some registers, an
>> interrupt handler and static information about the type of EDAC
>> device.
>
> Yeah, it's not that - I'm just getting worried that I'm receiving an
> EDAC driver for each piece of silicon out there and would like to still
> keep drivers/edac/ sane and be able to control that wild growth.
>
> I'm just thinking out loud here, bear with me pls:
>
> Frankly, having a single compilation unit contain similar silicon
> functionality could be a good way to put a hold on the growth but the
> disadvantage of this is fatter drivers. Which wouldn't matter all too
> much but after a certain level of fat, they might need splitting.
>
> And the highbank version is nothing but the big probe routine and a
> small irq handler.
>
> And the PL310 is similar but also with a poller.
>
> I guess, if they don't share functionality at all, putting them together
> might not be worth it. Hohummm.
>
> Thanks.
>
> --
> Regards/Gruss,
>     Boris.
>
> Sent from a fat crate under my desk. Formatting is fine.
> --
--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/